The vulnerability requires two key runtime behaviors: 1) Processing of attacker-controlled data from ML job indices, and 2) Output generation without proper escaping. The identified functions are core components of Elasticsearch's ML results pipeline:
- TransportGetResultsAction.doExecute is the entry point for results retrieval requests. It would appear in stack traces when users view ML job results.
- JobResultsProvider.processBuckets handles low-level data extraction from stored ML results. Both would handle raw attacker data before any escaping occurs.
While exact patch details are unavailable, the CVE description explicitly ties the vulnerability to ML job result viewing. These functions represent the most probable locations where unescaped data would flow from document storage to HTTP responses in affected versions. Confidence is medium due to reliance on architectural patterns rather than explicit patch analysis.